{"id":4724,"date":"2024-04-15T06:50:00","date_gmt":"2024-04-14T21:50:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=4724"},"modified":"2024-04-15T06:50:00","modified_gmt":"2024-04-14T21:50:00","slug":"eagle-finch-recurrentgemma","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=4724","title":{"rendered":"Eagle, Finch, RecurrentGemma"},"content":{"rendered":"\n<p>Transformer\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u4ee3\u308f\u308a\u3046\u308b\u30e2\u30c7\u30eb\u306b\u95a2\u3059\u308b\u5831\u544a\u304c\u51fa\u3066\u3044\u305f\u3002Eagle, Finch\u306fRWKV\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\uff08<a href=\"https:\/\/devneko.jp\/wordpress\/?p=4630\">DBRX, Jamba, Grok-1.5, RWKV Finch \u2013 arXiv\u6700\u65b0\u8ad6\u6587\u306e\u7d39\u4ecb (devneko.jp)<\/a>\u306a\u3069\uff09\u306e\u7814\u7a76\u6210\u679c\u3067\u975e\u5e38\u306b\u307e\u3068\u307e\u3063\u305f\u8ad6\u6587\u3001RecurentGemma\u306f<a href=\"https:\/\/devneko.jp\/wordpress\/?p=4503\">1 bit(1.58 bit)\u306aLLM\u3068HAWK\u30fbGriffin \u2013 arXiv\u6700\u65b0\u8ad6\u6587\u306e\u7d39\u4ecb (devneko.jp)<\/a>\u306eGriffin\u3092\u53d6\u308a\u5165\u308c\u305f\u30aa\u30fc\u30d7\u30f3\u306a\u30e2\u30c7\u30eb\u3067\u3042\u308b\u3002\u65b0\u305f\u306a\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u671f\u5f85\u3057\u305f\u3044\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Eagle and Finch: RWKV with Matrix-Valued States and Dynamic Recurrence\u00a0<\/strong>[37.0]<br>\u672c\u7a3f\u3067\u306f,RWKV(RWKV-4)\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u6539\u826f\u3057\u305f\u30b7\u30fc\u30b1\u30f3\u30b9\u30e2\u30c7\u30eb\u3067\u3042\u308bEagle(RWKV-5)\u3068Finch(RWKV-6)\u3092\u7d39\u4ecb\u3059\u308b\u3002 \u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u8a2d\u8a08\u306e\u9032\u6b69\u306b\u306f\u3001\u30de\u30eb\u30c1\u30d8\u30c3\u30c9\u884c\u5217\u5024\u72b6\u614b\u3068\u52d5\u7684\u518d\u5e30\u6a5f\u69cb\u304c\u542b\u307e\u308c\u308b\u3002 \u6211\u3005\u306f1.12\u5146\u306e\u30c8\u30fc\u30af\u30f3\u3092\u6301\u3064\u65b0\u3057\u3044\u591a\u8a00\u8a9e\u30b3\u30fc\u30d1\u30b9\u3068\u3001\u5f37\u5316\u3055\u308c\u305f\u591a\u8a00\u8a9e\u6027\u306e\u305f\u3081\u306egreedy\u30de\u30c3\u30c1\u30f3\u30b0\u306b\u57fa\u3065\u304f\u9ad8\u901f\u30c8\u30fc\u30af\u30f3\u5316\u5668\u3092\u5c0e\u5165\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2404.05892v2\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2404.05892v2\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 10 Apr 2024 19:34:38 GMT)<\/li>\n\n\n\n<li>RWKV\u306e\u6700\u65b0\u30d0\u30fc\u30b8\u30e7\u30f3\u306e\u8ad6\u6587\u3001\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u7d50\u679c\u3092\u898b\u308b\u9650\u308atransformer\u30d9\u30fc\u30b9\u306e\u6700\u65b0\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3068\u6bd4\u3079\u3066\u3082\u826f\u3044\u52dd\u8ca0\u306b\u306a\u3063\u3066\u304d\u3066\u3044\u308b\u3002\u5b66\u7fd2\u6642\u306e\u8a08\u7b97\u30b3\u30b9\u30c8\u3068\u6027\u80fd\u3067\u306fMamba\u3088\u308a\u3082\u30b3\u30b9\u30c8\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u304c\u3088\u3055\u305d\u3046\u3002<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f<a href=\"https:\/\/huggingface.co\/RWKV\">RWKV (RWKV) (huggingface.co)<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>RecurrentGemma: Moving Past Transformers for Efficient Open Language Models\u00a0<\/strong>[103.6]<br>\u672c\u7a3f\u3067\u306f,Google\u306e\u65b0\u3057\u3044Griffin\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u7528\u3044\u305f\u30aa\u30fc\u30d7\u30f3\u8a00\u8a9e\u30e2\u30c7\u30eb\u3067\u3042\u308bRecurrentGemma\u3092\u7d39\u4ecb\u3059\u308b\u3002 Griffin\u306f\u3001\u8a00\u8a9e\u306b\u304a\u3051\u308b\u512a\u308c\u305f\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u9054\u6210\u3059\u308b\u305f\u3081\u306b\u3001\u7dda\u5f62\u53cd\u5fa9\u3068\u5c40\u6240\u7684\u306a\u6ce8\u610f\u3092\u7d44\u307f\u5408\u308f\u305b\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2404.07839v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2404.07839v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 11 Apr 2024 15:27:22 GMT)<\/li>\n\n\n\n<li>\u3053\u3061\u3089\u306fGriffin\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u3088\u308b\u30aa\u30fc\u30d7\u30f3\u30e2\u30c7\u30eb\u30022B\u3067\u6bd4\u8f03\u3057\u3066Gemma\u3068\u307b\u307c\u540c\u6027\u80fd\u3001\u30b9\u30eb\u30fc\u30d7\u30c3\u30c8\u306f\u5927\u5e45\u306b\u5411\u4e0a\u3057\u3066\u3044\u308b\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/google-deepmind\/recurrentgemma\">google-deepmind\/recurrentgemma: Open weights language model from Google DeepMind, based on Griffin. (github.com)<\/a>\u3001\u30e2\u30c7\u30eb\u306fKaggle\u3067\u516c\u958b\u3055\u308c\u3066\u3044\u308b\u3002<a href=\"https:\/\/www.kaggle.com\/models\/google\/recurrentgemma\">RecurrentGemma | Kaggle<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Transformer\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u4ee3\u308f\u308a\u3046\u308b\u30e2\u30c7\u30eb\u306b\u95a2\u3059\u308b\u5831\u544a\u304c\u51fa\u3066\u3044\u305f\u3002Eagle, Finch\u306fRWKV\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\uff08DBRX, Jamba, Grok-1.5, RWKV Finch \u2013 arXiv\u6700\u65b0\u8ad6\u6587\u306e\u7d39 &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=4724\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;Eagle, Finch, RecurrentGemma&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[223,293,346],"class_list":["post-4724","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-oss","tag-rwkv"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4724","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4724"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/4724\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4724"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4724"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4724"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}